Studying sampling effects in MODIS and MISR aerosol data via GOCART model data

Successful Aerosol Optical Depth (AOD) retrievals are sampled unevenly in space and time due to several reasons. Different satellites observe aerosol at different times of a day, the swath width lead to differences in coverage, some bright surfaces do not allow a reliable estimation of AOD, clouds prevent retrieval of AOD, etc. Since the data gaps due to sampling are far from random, we can expect sampling to result in systematic biases. To make things worse, the gaps in data due to clouds correlate with the AOD itself. We evaluate the magnitude of possible systematic effects of spatial-temporal sampling on AOD measurements by different instruments by employing the Goddard Chemistry Aerosol Radiation and Transport (GOCART) model data. Contrary to various previous efforts, our approach interpolates smoothly varying model data to space-time locations of Level 2 AOD data measured by each of the instrument. We compute AOD zonal means using GOCART data (a) at the original model uniform grid and (b) only at locations of successful AOD retrievals from three instruments: MODIS-Aqua, MODIS-Terra, and MISR. We analyze the differences in zonal means and in the seasonal AOD variations due to sampling of GOCART data that correspond to sampling of satellite data. Effects of model resolution are also addressed. Analysis of differences due to sampling in smoothly varying model data allows us to set a lower limit of sampling effects on zonal mean, the amplitude of seasonal variations, and other parameters derived from unevenly sampled satellite data.